The attic air hung thick and dusty, catching in Leo’s throat. He coughed, waving a hand in front of his face. Sunlight sliced through a gap in the boarded-up window, illuminating swirling motes of dust that danced around the cardboard box he’d just dragged out. It was his mother's.

He knelt, slowly lifting the lid. Inside, meticulously organized, were stacks of papers. Spelling tests, art projects with crayon scribbles, book reports on topics he barely remembered. Then he saw his name, scrawled in a familiar, childish hand: “Leo.”

His stomach churned, a knot tightening in his chest. He ran a hand through his hair, the cardboard scratching against his knuckles. He remembered, vividly, the tantrums. The slammed doors. The cruel words. All because of a feeling he didn't name then, but knew now.

He sank back on his heels, a sudden weariness washing over him. He closed the box, a feeling of unease settling into the pit of his stomach.

Emotion: guilty

Cluster: Shame / Guilt
PC1 (Valence): -1.62 Negative
PC2 (Disposition): 1.23

Role in Research

This story is one of 1,000 stories generated for the emotion guilty. During extraction, it was fed through Gemma4-31B and its hidden state activations were captured at 11 layers.

The mean activation across all 1,000 guilty stories, after denoising with neutral dialogue baselines, produces the guilty emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the guilty vector is projected through the unembedding matrix.

Promoted:
S0.492
L0.398
0.354
been0.323
ashamed0.319
Suppressed:
de-0.400
(!)-0.279
la-0.276
😍-0.274
"!-0.264